The Ptolemaic era (305–30 BCE) is an important period of Ancient Egyptian
history known for its material and scientific advances, but also intermittent political and social unrest in the form of (sometimes widespread) revolts against the Ptolemaic elites. While the role of
environmental pressures has long been overlooked in this period of Egyptian
history, ice-core-based volcanic histories have identified the period as
experiencing multiple notable eruptions, and a repeated temporal association between explosive volcanism and revolt has recently been noted. Here we analyze the global and regional (Nile River basin) hydroclimatic response to a unique historical sequence of four large and closely timed volcanic eruptions (first a tropical one, followed by three extratropical northern hemispheric events) between 168 and 158 BCE, a particularly troubled period in Ptolemaic history for which we now provide a more detailed hydroclimatic context. The NASA (National Aeronautics and Space Administration) GISS (Goddard Institute for Space Studies) ModelE2.1 Earth system model simulates a strong radiative response with a radiative forcing (top of atmosphere) of
Explosive volcanic eruptions that result in high-altitude sulfate aerosol
distribution across one or both hemispheres can diminish insolation with global and regional climatic impacts (e.g., Robock, 2000; Toohey et al., 2019). The sulfate aerosols resulting from (for example) the 1991 Mt. Pinatubo eruption of
Explosive volcanic eruptions represent the major natural source of forced variability in the climate system at yearly to decadal timescales (Schmidt et al., 2011; Colose et al., 2016; Swingeduow et al., 2017; Khodri et al., 2017). Powerful explosive eruptions can inject sulfur-rich gases into the stratosphere, where they oxidize to form sulfate aerosols that can persist for months to years, impacting climate on regional to global scales. Volcanic stratospheric aerosols can cause troposphere cooling by scattering incoming shortwave radiation while also heating the stratosphere (Robock and Mao, 1992). Unequal north–south stratospheric heating due to volcanic aerosols concentrated in lower latitudes after tropical eruptions can influence major modes of circulation and surface climate variability, such as the Arctic Oscillation/North Annular Mode (AO/NAM) and North Atlantic Oscillation (NAO), driving an enhanced westerly airflow (Shindell, 2004; Christiansen, 2008; Zanchettin et al., 2022). The post-volcanic surface temperature response can affect the El Niño/La Niña phase and Southern Oscillation, and have a long-term impact on Atlantic Meridional Overturning Circulation (AMOC) strength (Khodri et al., 2017; Wahl et al., 2014; Robock and Mao, 1995; Pausata et al., 2015). Extratropical eruptions are usually thought to have a weaker impact than tropical eruptions. This arises in part because of the background Brewer–Dobson circulation upwelling in the tropics and downwelling at higher latitudes, which affects the stratospheric lifetime of volcanic aerosols (Kirtman et al., 2013; Myhre et al., 2013; Schneider et al., 2009). Recent studies have, however, illustrated the potential for a dynamically induced and disproportionally strong climate forcing from such eruptions (Toohey et al., 2019).
Volcanic injection of sulfur-containing compounds can, too, influence stratospheric chemistry, yielding further complex atmospheric and climatic responses upon interacting with water and halogens (LeGrande et al., 2016; Brenna et al., 2020; Staunton-Sykes et al., 2021). Paleoclimate records and climate modeling suggest that the dynamical response to volcanic aerosol causes a net (but regionally variable) drying and impacts global rainfall patterns (PAGES Hydro2k Consortium, 2017; Colose et al., 2016; Liu et al., 2016; Iles and Hegerl, 2014). Trenberth and Dai (2007) thus analyzed the impact of the Pinatubo (1991) eruption on terrestrial precipitation and river streamflow and found an increase in associated drought conditions in 1992. Joseph and Zeng (2011) suggested that volcanically induced rainfall anomalies over land and ocean can seasonally modulate tropical drought. Hemispheric biases in volcanic aerosol distribution can, moreover, impact the movement of ITCZ by constraining its summertime migration into the energetically deficit hemisphere (Colose et al., 2016; Xian and Miller, 2008). Effectively, the ITCZ shifts “away” from the hemisphere with the greatest aerosol burden. For tropical eruptions, even those producing roughly even hemispheric burdens, this “movement” is typically more southward owing to the larger amount of land in the Northern Hemisphere and the greater ocean area (and higher thermal capacity) of the Southern Hemisphere.
For Africa, eruptions producing asymmetrical latitudinal aerosol burdens
(e.g., Katmai in 1912, El Chichón in 1982) may have enhanced 20th century Sahelian droughts by influencing the strength and position of Hadley cells (Haywood et al., 2013). Of the Nile, monsoon rainfall over the Ethiopian highlands contributes (mainly via the Blue Nile and Atbara River)
One of the most richly documented periods of ancient Egyptian history is the Ptolemaic era, 305–30 BCE, with Egypt ruled by Greeks in a lineage beginning with Ptolemy I Soter (d. 283 BCE), previously one of Alexander the Great's key generals and instrumental in Egypt's conquest. The period distinguishes itself for mixing Greek and Egyptian traditions, its material, cultural, and scientific achievements (e.g., the founding of Alexandria with its Great Library), but also its chronic political instability (McGing, 1997; Ludlow and Manning, 2016, 2021). External environmental influences have been little considered in this, despite the dependence of Egyptian agriculture on the summer flood. However, recent work has revealed a repeated close coincidence in the timing of many (if certainly not all) internal revolts and ice-core-based dates of inferred-tropical and NH extratropical eruptions, that appear statistically significant (i.e., non-random) (Ludlow and Manning, 2016, 2021; Manning et al., 2017). One example is the “Great Theban Revolt” starting in c. 207 BCE, shortly after a 209 BCE tropical eruption (Sigl et al., 2015), with extensive territories lost to two native Pharaohs (Ludlow and Manning, 2016; Ludlow et al., 2023).
That the revolt and eruption dates under study derive from independent
chronologies (documentary and ice core) helps exclude potential biases in
estimating this statistical significance. For example, inflated correlations
may result between events known from the same sources (e.g., between extreme
weather and societal stresses such as famine, if those instances of extreme
weather that contributed to such stresses were more likely to have been
documented than those that did not; White and Pei, 2020). It is a truism
that correlation does not establish causation. Genuine causality is, however, implied where significance testing suggests an observed correlation is unlikely to have arisen randomly, though this does not determine the direction or character of causality (Izdebski et al., 2022). Statistical
significance may, however, be sensitive to many factors. These include here (1) the choice of statistical test, (2) the choice of revolt and eruption dates (if uncertainties exist), (3) judgments as to what constitutes “revolt” (vs. phenomena like food riots motivated more by desperation than politics), and (4) judgments concerning which eruptions to include in testing (e.g., seeking those with a meaningful impact vs. non-climatically effective eruptions introducing “noise” into the analysis), assessed by estimated volumes and heights of atmospheric
Logic dictates that the direction of any genuine causality must flow here from eruption to revolt (Izdebski et al., 2022). Further confidence in its reality arises from the existence of plausible mechanisms connecting volcanic hydroclimatic variability with revolt (i.e., via reduction of the Nile summer flood and consequent societal impacts). Much work remains to further characterize this causality, how direct or indirect it may have been, and whether this changed meaningfully through time (and between revolts that varied in geography and scale) according to (or in interaction with) other coincident potential causes (from longer-term developments promoting chronic vulnerabilities, to more acute political and socioeconomic stresses). White and Pei (2020) argue that such questions represent a key challenge for climate historians and related scholars, recommending that potential causes are assessed using a framework of necessary and sufficient conditions (put simply; see also Ludlow et al., 2023). Gao et al. (2021) employ an alternative framework (though not mutually exclusive to that recommended by White and Pei, 2020), wherein the role of volcanically induced hydroclimatic “shocks” in the collapse of Chinese dynasties is characterized along a spectrum of causality in which smaller hydroclimatic shocks might trigger collapse when enabled or facilitated by higher pre-existing societal stress (i.e., warfare), while larger shocks might act with greater independence, triggering collapse even with less substantial pre-existing stress.
An alternative framing in many climate–conflict studies (again not incompatible with that proposed by White and Pei, 2020 or employed by Gao et al., 2021) is to delineate multiple identifiable “pathways” that may enable or lead (through material (e.g., economic), political, cultural, or psychological channels) to links between hydroclimatic variability and various forms of conflict (see Hsiang and Burke, 2014 and Ide, 2017 for reviews). Success here requires that statistical findings are interpreted with due reference to the relevant historical, political, and geographical contexts. For Ptolemaic Egypt, one hypothesized pathway involves the societal impacts and responses to sudden hydroclimatic variability that can follow explosive volcanism and influence (alongside other regional factors) the intensity of the African monsoon. When this causes a “failure” of the Nile summer flood, many adverse societal impacts may follow. These include harvest failure (seen also in other periods of Egyptian history; e.g., Hassan, 1997a, b; Mikhail, 2015), potentially prompting subsistence-driven migration to urban areas, with inability to meet state taxation demands (payable in grain) also potentially necessitating the sale of hereditary familial lands (Manning, 2003; Manning et al., 2017). These stressors might work in tandem with the psychological and religious significance of a “failed” Nile flood, something widely feared, and which could be interpreted (even propagandized to foment revolt) as signaling divine displeasure at the Pharaoh (Ludlow and Manning, 2021; Ludlow et al., 2023). In the Ptolemaic context, when some native Egyptian elites were likely resentful of Greek rule, with taxation and other advantages given to those of Greek backgrounds (McGing, 1997; Ludlow et al., 2023), a Nile failure may have held particular political potency.
Huhtamaa et al. (2022) have called for case studies of the hydroclimatic and socioeconomic impacts of specific eruptions to advance understandings of human–environmental causalities. For Ptolemaic Egypt, the 160s BCE are particularly relevant in experiencing considerable internal revolt and instability. Indeed, the Ptolemaic dynasty might have fallen here without self-interested Roman intervention against the Seleukid empire (rivals to the Ptolemies) after their invasion (170–168 BCE) of Egypt under King Antiochus IV (Grainger, 2010; Blouin, 2014; Manning et al., 2017). This decade is also remarkable for three notable volcanic eruptions (168, 164, 161 BCE), with another in 158 BCE (Sigl et al., 2015), which we term the “eruption quartet”. Substantial sulfate across both poles identifies the first (168 BCE) as the largest and likely tropical, followed by three moderate eruptions in the Northern Hemisphere extratropics (Sigl et al., 2015). While high-resolution paleoenvironmental proxies for Egypt are effectively absent for this period, the hydroclimatic impacts of these eruptions can be explored by climate modeling.
Few studies have examined the climatic–societal effects of eruption clusters. These include the cluster between 1108 and 1110 CE (Guillet et al., 2020), the “double event” of the 6th century in 536 and 540 CE (Toohey et al., 2016), and the cluster from 1637 to 1641 (Stoffel et al., 2022; Huhtamaa et al., 2022). These studies variously employed paleoclimatic data, written evidence, and/or climate modeling to reveal strong negative post-eruption temperature anomalies for the Northern Hemisphere, thereby suggesting the potential for diminished crop yields and providing a climatic context to better understand the human history of these periods. Here, we intend to advance our understanding of the likely hydroclimatic impact of the 168–158 BCE eruption quartet as a foundation for ongoing efforts to more securely establish and qualify the causality underlying the observed association between eruptions, Ptolemaic-era revolts, and other political and socioeconomic phenomena and developments.
We thus use a computationally expensive but more sophisticated version of the National Aeronautics and Space Administration (NASA), Goddard Institute for Space Studies (GISS) Earth system model, GISS ModelE2.1-MATRIX (Bauer et al., 2008, 2020), to simulate the 168–158 BCE eruption quartet and regional hydroclimate responses over the Nile River basin. Section 2 presents model details and experiment methodology. Estimation of background climate for the 2.5 ka period (orbital and greenhouse gas (GHG) changes), alongside impacts due to PMIP4 (Paleoclimate Model Intercomparison Project, phase 4) vegetation cover estimates for the period are considered in Sect. 3. Particular subsections evaluate the GISS ModelE for its capability to resolve the microphysical properties of volcanic aerosols during this period and to analyze the radiative impacts of the aerosols from these eruptions, which control their radiative and hydroclimatic impacts (Timmreck et al., 2009, 2010; Schmidt et al., 2010). The discussion and conclusion (Sect. 4) summarizes our results and considers how they can advance understandings of the period's fraught history in Egypt.
We used the NINT (Non-INTeractive) version (Kelley et al., 2020) of GISS ModelE2.1 to simulate background climate conditions corresponding to 2500 years before present (2.5 ka, kilo-years BP), similar to protocols for the mid-Holocene (6ka) coordinated experiment (Kageyama et al., 2017), but adjusting trace gases and orbital forcing for 2.5 ka. The term “non-interactive” means that atmospheric composition and climate are decoupled, so any changes in composition are handled by external input only. Once our model attained an equilibrium climate state, we enabled atmospheric composition–climate interactions for our experiments, described below.
GISS ModelE2.1 is a state-of-the-art Earth system model contributing to the
Climate Model Intercomparison Project (CMIP) Phase 6 (Eyring et al., 2016).
The model's atmospheric component simulates on a horizontal resolution of
2
Approximate eruption locations are crucial to estimate climatic impacts
(Toohey et al., 2016; AGU Poster:
A control simulation for the 2.5 ka period was performed using the PMIP Phase 4 protocols for the mid-Holocene (6 ka) experiment, altered for conditions appropriate to 2.5 ka. This included altering the orbital forcing, greenhouse gases (
Figure S1 in the Supplement shows the major vegetation plant function type (PFT) cover changes under the PMIP4 sensitivity vegetation protocols after linearly interpolating for the 2.5 ka period. The 2.5 ka equilibrated simulation with MATRIX was then extended for 70 more years with a corrected dust tuning (a typical process when equilibrating the model on a new climate state), and a further 130 years with the linearly interpolated PMIP4 vegetation described above (see Table S1 in the Supplement for details of control runs and annual global mean time series of surface air temperature and precipitation in Fig. S2). This run equilibrated very quickly, and no further tuning was needed. We thus used the last 100 of the total 130 years of that equilibrated run as the base climate for our analysis. An ensemble of 10 members with active volcanic eruptions was simulated using a restart file every 10 years during the last 100 years of the control simulation, corresponding to 2.5 ka period as summarized in Table S1, following the same approach as performed for the CMIP6 ensemble simulations (Kelley et al., 2020).
The starting time point for each ensemble member is shown by blue vertical lines in Fig. S2. Each member started 1 January of the year 169 BCE and ran for 16 years, with each eruption happening on 15 June of the 2nd, 6th, 9th, and 12th years modeled. Because exact eruption dates cannot be determined from ice-core sulfate deposition data, due to ice-core chronological uncertainties and variable lags between eruptions and the deposition of sulfate in the ice, we selected a summer eruption date to investigate impacts on Northern Hemisphere monsoon and wintertime atmospheric circulation. We also note that modeling accuracy will depend partly upon the accuracy of the ice-core-based forcing reconstruction employed. Uncertainties can arise, for example, because of variation in sulfate deposition across the polar regions for any given eruption. It is thus notable that Sigl et al. (2015) employed several Antarctic and Greenland ice cores in their reconstruction, helping to average out regional variability in deposition, but our results can be revisited as reconstructions incorporate more ice cores.
We first evaluated the 2.5 ka control run for the purposes of providing a precise background climate to investigate the impacts of the 168–158 BCE volcanic quartet.
Seasonal mean (annual, DJF, and JJAS) surface air temperature (top
row) for the 2.5 ka period equilibrium run, differences from the
preindustrial period (2.5 ka
Mean surface air temperature for annual, DJF, and JJAS seasons (top
row) and seasonal mean precipitation (third row from top) for the equilibrium runs with the PMIP4 vegetation for the 2.5 ka period, and surface temperature difference (second row from top) plus the seasonal precipitation differences (bottom row) for the 2.5 ka period, as simulated by GISS ModelE2.1. We have used a short initial notation for our forcings to denote the difference (ORB
We compared the 2.5 ka equilibrium climate with only GHG, orbital, and non-anthropogenic forcing changes against a preindustrial (year 1850) control run to evaluate the impact (alone) of orbital and greenhouse gas changes on our base climate state. Surface air temperatures showed globally minimal differences, with a warming of Northern Hemisphere high latitudes due to the different orbital forcing for all seasons (Fig. 1). The implications of changes in orbital forcing for 2.5 ka are thus evident in the surface temperature, but the Northern Hemisphere monsoon season (JJAS) and winter season (DJF) rainfall was slightly reduced along the Northern Equatorial Belt. This points to the limitation of the GISS model in not having an interactively dynamic vegetation component to reproduce the known mid-Holocene wet African land cover conditions (Harrison et al., 2015; Tiwari et al., 2022). Numerous studies have demonstrated that including bio-geophysical feedbacks and atmospheric dynamics helps to successfully model the wet African conditions for mid-Holocene (Kutzbach et al., 1996; Claussen et al., 2003; Kutzbach and Liu, 1997; Hewitt and Mitchell, 1998). Using PMIP4 vegetation over northern hemispheric regions is known to provide a solution to the same long-standing issue with CMIP3/CMIP5 models that fail to reproduce these wet African conditions for the mid-Holocene (Harrison et al., 2015).
The comparison of mean climate for the 2.5 ka period for inclusion of PMIP4 vegetation is shown in Fig. 2, for the mean surface air temperature and precipitation for the annual, DJF, and Northern Hemisphere monsoon (JJAS) seasons.
GISS ModelE2.1 simulated a global mean surface air temperature (SAT) of 14.4, 12.8, and 15.7
Annual zonal mean of ground albedo
We also analyzed zonal changes of longwave and shortwave radiation at the
top of the atmosphere with our altered surface albedo (Fig. 3).
Vegetation–albedo feedback from the inclusion of woody forest over higher
latitudes and shrubs and steppes over northern Africa is important in the
additional monsoon season rainfall seen for North Africa. Greater vegetation
cover for the Sahara and at higher northern hemispheric latitudes alters
surface albedo by
Details of eruptions applied in this experiment, with each eruption happening on 15 June of the 2nd, 6th, 9th, and 12th model years.
We simulated a series of four eruptions occurring mid-June during the 2nd, 6th, 9th, and 12th years of our runs (as per Sect. 2.2 and Table 1). Explosively injected
Monthly mean top of the atmosphere radiative balance perturbation due to volcanic aerosols for the entire simulation length. Orange/red shows the longwave radiative response, light/dark blue represents the shortwave, and gray/black represents the net (TOA) radiative change averaged at the global scale. The light-colored solid lines represent individual ensemble members, and the dark-colored lines show the ensemble mean. The green vertical dashed lines show when the eruptions happened.
With 22.5 Tg of
Globally averaged changes in microwave sounding unit (MSU) temperature for the lower stratosphere (TLS)
Figure 5a shows the monthly change in microwave sounding unit (MSU) temperature for the lower stratosphere (TLS) as calculated by the model, which is a typical metric for present-day evaluation of modeled stratospheric temperatures against satellite data. It covers the lower stratosphere, where volcanic aerosols mostly reside, and represents the
local atmospheric response of longwave absorption by them. After Pinatubo (1991), a lower stratospheric warming on the order of 2–3
Figure 5c presents the aerosol optical depth (AOD), a measure of atmospheric opacity to incoming radiation calculated as the extinction (sum of scattering and absorption) of shortwave radiation at the 550 nm wavelength. The model simulated an AOD anomaly of around 0.21 for the first 18 months after E1, decreasing as aerosols were progressively removed. The subsequent eruptions produced an AOD of the order of
In the upper troposphere and lower stratosphere, the impact of each eruption
was distinct with a near-complete recovery to background AOD levels observed
after each (i.e., before the next) event; however, at the surface, a lag in
recovery time was evident (Fig. 5b). The net impact of radiative flux perturbations following the eruptions is summarized using the global surface air temperature change for the entire period. The model produced a mean cooling of
Radiative forcing from volcanic aerosols is tightly controlled by aerosol
size (Lacis et al., 1992; Hansen et al., 1980). The aerosol effective radius,
Time series of global ensemble mean vertical profile of sulfate
aerosol
The aerosol extinction vertical profile (Fig. S5a) shows that the radiative
impact of the E1 tropical eruption in the lower stratosphere was prolonged
as compared to the later extratropical eruptions. Lower stratospheric heating affects the dynamics of the stratosphere; after tropical eruptions enhanced tropical upwelling and extratropical downwelling within the atmospheric circulation impact upon the transportation of trace species such as ozone (
Hovmöller-type plot showing the zonally averaged temporal
dispersion of volcanic aerosols in terms of AOD change at 550 nm
The Hovmöller-type plot (Fig. 7a and b) shows the differences between the
zonally averaged AOD at 550 nm and surface air temperature response for the
ensemble means of the volcanic eruption simulations compared to the mean
climatology of the control simulation. The statistical significance level is
estimated using the two-tail Student
An observed lag of more than 12 months in peak surface temperature response
(Fig. 5b) after E1 correlates well with the modeled aerosol distribution and is consistent with reporting for similar events (e.g., Jungclaus et al., 2010; Klocke, 2011). The peak global mean surface temperature response thus appeared when aerosols from the tropical eruption (E1) had extended across the Northern Hemisphere extratropics and polar regions. It should be noted that northern extratropical land surfaces
responded quickly to the attenuated post-eruption shortwave radiative flux
compared to the tropics. The zonally averaged surface temperature response
(Fig. 7b) showed that a strong cooling of 1.0–1.5
The seasonality of surface temperature response revealed a more substantial cooling during the boreal summer season for all four eruptions and for E1 also revealed the expected post-tropical-eruption winter warming pattern over Europe. Figure S6 shows the spatial pattern of the surface temperature response to volcanic aerosols over the four seasons directly following E1 (JJA and SON for the eruption year and DJF and MAM for first year following). The response for the first two seasons was confined to the tropics, but moved to higher latitudes thereafter. As evident in Fig. S6, the winter (DJF) post-eruption warming over Europe (not statistically significant) and an observed cooling over North America may result from the same fundamental atmospheric dynamics noticed after Pinatubo (1991) (Robock, 2000; Robock and Mao, 1992).
The global lower stratospheric temperature response in terms of MSU TLS data
was discussed in Sect. 3.2. Interestingly, Fig. 7c shows that the latitudinal anomaly of the lower stratosphere warming was centered along the Equator and largely constrained between 60
We used a relatively coarse resolution Earth system model having a simplified parameterization that is skilled in simulating the large-scale patterns of climate response to natural and anthropogenic forcings (Kelley et al., 2020; Miller et al., 2021; Nazarenko et al., 2022). Studies of observational records plus modeling efforts have demonstrated that the cascading impact of an altered radiative balance at the top of the atmosphere due to volcanic eruptions is reflected in the hydrological cycle in regional patterns of seasonal rainfall change (e.g., Robock and Liu, 1994; Robock, 2000; Trenberth and Dai, 2007; Schneider et al., 2009; Iles et al., 2013; Iles and Hegerl, 2014; Timmreck, 2012). We investigated the hydrological cycle response to the 168–158 BCE eruption quartet at both a global and regional scale, paying particular attention to the northern hemispherical monsoon season (JJAS) for the first 2 years following each eruption. Any individual ensemble member might best represent the historical reality, but it is impossible to select the most accurate member absent supporting observational data from the period. Also, added noise due to natural variability can be greater at the regional scale, even to the extent of altering the sign of observed changes among the individual ensemble runs. Thus, we mainly focused upon the mean from across the ensemble when examining the response to the eruptions for the various climate variables considered. Figure 8 shows the Hovmöller-type plot of the zonal mean precipitation anomaly relative to the annual cycle climatology of the 100-year-long control simulation.
Hovmöller-type plot showing the zonally averaged rainfall anomaly for the entire period as the spatiotemporal response of global rainfall to our series of volcanic eruptions. Circled cross marks show the locations and timing of the eruptions. Black dots point out regions where changes are not statistically significant at the 95 % confidence level.
The ensemble zonal mean post-eruption rainfall change showed a substantial
negative trend in the Northern Hemisphere due to the volcanic aerosol-induced cooling. A robust negative anomaly on the order of 0.3–0.4 mm d
Mean change (mm d
We further evaluated the spatial patterns of change in mean rainfall during
the Northern Hemisphere monsoon season (JJAS) (Fig. 9). We averaged the three monsoon seasons (eruption year and next 2 years) after the more potent tropical eruption (E1) and two monsoon seasons (eruption year and next year)
after each of the extratropical eruptions (E2, E3, and E4), focusing principally on statistically significant responses. After E1, summer monsoon
rainfall appeared strongly suppressed over many major Northern Hemisphere
monsoon regions. Importantly for our focus on Egypt, African monsoon
rainfall showed a notable decrease of 0.5–1.0 mm d
Ensemble mean rainfall difference from the climatological control for each of the first three monsoon seasons (JJAS; rows) after each eruption (columns) over equatorial and northern Africa. The blue boundary line shows the present-day Nile River basin, broadly similar to the river extent approximately 2.5 ka years ago. The red stippling indicates regions over which change in rainfall is not statistically significant at a 95 % confidence level.
Similar patterns of suppressed boreal monsoon season rainfall were observed
following the extratropical eruptions (E2–E4), but a particularly notable
east–west band over land and ocean (broadly confined between slightly north
of the Equator and 30
Our modeling suggests that all four eruptions, 168–158 BCE, are likely to
have influenced rainfall over different monsoon regions in the Northern
Hemisphere for 2–3 years after each eruption, in combination producing a
sustained deficit for more than a decade. Focusing on the North African
monsoon region, Fig. 10 shows 3 consecutive years of JJAS rainfall over
equatorial and northern Africa (encompassing the Nile Basin) after each eruption. The African monsoon exhibited notably reduced rainfall of
Annual river flow anomaly (km
Annual mean change (%) and standard deviation in water mass flow
over the Nile River catchment for 3 consecutive years after each eruption.
Control run variability (interannual standard deviation about the decadal
mean,
Spatial patterns in total cloud cover (Fig. S7) for the three consecutive
post-eruption monsoon seasons showed a decrease of up to 10 % over East
Africa and the adjacent Indian Ocean region. This is consistent with the
above-reported negative rainfall anomalies (
Monthly time series of individual ensemble and mean of surface
temperature response (
We also analyzed the mass of total annual water flow averaged over the Nile
River basin (blue line, Fig. 11) as representative of Nile flooding and
discharge at the river's mouth to summarize the volcanic impacts on Nile
flooding. We used the ArcGIS shape file of the modern Nile River basin (blue
line in Fig. 10) to generate the weights for a fraction of the model grid
cell over the basin boundary and to quantify the amount of water flow in the
river basin. Table 2 presents the percentage annual deficit (
The spatial patterning of response across a basin as complex as the Nile is
a critical consideration (Fig. 11). After E1, the above-described rainfall
suppression is associated with a notable reduction in annual river flow
observed over effectively the entire river basin, with a simulated decrease of approximately 30, 40, and 15 km
Annual Nile River flow changes averaged over the northern (red) and southern (blue) parts of the basin (divided at 10
To summarize the hydroclimatic impact of the volcanic quartet on the Nile River basin, Fig. 12a shows that the Northern Hemisphere experienced a substantial cooling of
It is evident that the mean surface temperature response in the Northern
Hemisphere is significant at the control period's
These results are consistent with our earlier-described results (e.g., spatial rainfall variability over the Nile River basin, as per Figs. 10 and 11) and proposed mechanisms, alongside expectations from the literature (e.g., Manning et al., 2017). Thus, tropical eruptions (like E1) may be expected to produce a more consistent (negative) north–south flow response due to their more even interhemispheric aerosol burden and associated radiative impact. Extratropical NH eruptions (like E2–E4) that can result in a more asymmetric hemispheric aerosol burden may, by contrast, be expected to introduce contrasting flow anomalies by suppressing the northward migration of the ITCZ, negatively impacting flow in the Blue Nile and Atbara rivers by diminishing monsoon rainfall in the Ethiopian highlands, while potentially enhancing flow in the White Nile, fed by rainfall over the equatorial lakes.
Recent years have seen increasing interest in the role of hydroclimatic variability in human history, including by interdisciplinary teams combining evidence and methods across disciplinary divides (e.g., McCormick, 2011, 2019; Manning et al., 2017; Ludlow and Travis, 2019; van Bavel et al., 2019; Campbell and Ludlow, 2020; Degroot et al., 2021; Gao et al., 2021; Ljungqvist et al., 2021; Travis et al., 2022; Izdebski et al., 2022; Ludlow et al., 2023). For the pre-modern era, when systematic observations of hydroclimate become scarce, this effort depends increasingly upon natural archives (paleoclimatic proxies) that track variability at spatial and temporal resolutions sufficiently high to convincingly identify associations with societal phenomena (e.g., subsistence crises, migration, conflict), economic and demographic processes, and major historical events (e.g., “collapse” of empires). Work such as that by PAGES2k Network members offering paleoclimatic reconstructions and data collections (e.g., PAGES 2k Consortium, 2013, 2017) are thus crucial, although here the exclusive focus on the past 2000 years (for some proxies an artificial horizon and for others an aspiration), excludes many foundational periods and events in human civilization. This includes the development of advanced ancient societies in Asia, the Near East, and Mediterranean that are well documented and offer considerable potential for the study of socio-ecological systems.
Important work has still been possible using speleothems, sedimentary, and other archives (e.g., Drake, 2012; Schneider and Adali, 2014; Knapp and Manning, 2016; Sołtysiak, 2016), but there is often little direct temporal and/or geographical overlap between these early ancient world regions of rich human documentation and proxies (e.g., tree-ring-based) with precision and accuracy at annual-or-better resolutions. A major advance has been the publication of a chronologically precise and accurate bipolar ice-core-based volcanic forcing reconstruction for the past 2500 years (Sigl et al., 2015; Toohey and Sigl, 2017). The potentially global hydroclimatic impacts of major explosive eruptions makes this record widely geographically relevant, while the repeated incidence of major eruptions that can be detected through sulfate deposition in the polar ice sheets has allowed their use as “tests” of societal vulnerability and response to sudden hydroclimatic shocks, in both a statistical manner (e.g., Manning et al., 2017; Campbell and Ludlow, 2020; Gao et al., 2021; Ludlow et al., 2023) and in a complementary qualitative manner as “revelatory crises” (for this concept, see Solway, 1994 and Dove, 2014), in which tensions and vulnerabilities in political and economic systems are potentially exposed under pressure from sudden environmental variability (e.g., Ludlow and Crampsie, 2019; Huhtamaa et al., 2022; Ludlow et al., 2023).
For historical eruptions to act as tests in this manner or to be studied as potential “revelatory” crises, knowledge of their dating alone is insufficient, particularly given the regional and seasonal variability of volcanic hydroclimatic impacts and the sensitivity of these impacts to multiple variables such as the location, season, chemical composition, and height attained by volcanic ejecta (Robock, 2000; Cole-Dai, 2010; Ludlow et al., 2013). Even where instrumental or natural archives are available, but especially where these are thin or absent, climate modeling can provide insights into the expected impacts for particular regions, seasons, and related physical (e.g., riverine) systems (e.g., Toohey et al., 2016; McConnell et al., 2020; Mackay et al., 2022). This is true for modeling of idealized eruptions, but potentially even more so for models that produce “historical realizations” based upon actual forcing reconstructions (e.g., Tardif et al., 2019).
In this context, we have presented a modeling effort that explores the
impacts of a unique eruption quartet during the (historically tumultuous)
decade 168–158 BCE, with a focus on the Nile River basin. These target years are intermediate between the mid-Holocene and end of the preindustrial
periods, and representative background climate conditions are necessary to
investigate the climatic impact of such a short-term forcing (Zanchettin et
al., 2013). PMIP4 vegetation distributions (linearly interpolated for the
2.5 ka period from the mid-Holocene (Otto-Bliesner et al., 2017) to the end of the preindustrial period (taken as 1850) for the GISS ModelE2.1 (MATRIX) version (Kelley et al., 2020; Bauer et al., 2020) were therefore used to improve GCM simulations without a fully dynamic vegetation implementation (Harrison et al., 2015). Vegetation–albedo feedbacks due to greater prevalence of arid shrubs/steppe over Africa and of boreal forests over high latitudes were observed to induce a northward movement of the ITCZ over Africa (Sahara region) promoting a simulated rainfall increase of the order of 0.5–1.0 mm d
The GISS ModelE2.1 simulated a strong shortwave and longwave global radiative forcing of
The global hydrological cycle responded vigorously to the volcanically induced surface cooling in the GISS ModelE2.1, with a
For the equatorial and northern African landmass specifically, the GISS
ModelE2.1 produced a notable suppression of monsoon (JJAS) rainfall for all
eruptions, E1–E4. The onset of this response can be observed in the JJAS
season beginning with each eruption year itself, though the timing of the peak intensity and/or greatest spatial extent of this suppression varied (e.g., for E1 the greatest extent and peak intensity occurred for JJAS in
year 0, while for E2–E3 the peak intensity and greatest extent occurred in
year 1, and for E4 in year 0). The suppression centered (for all eruptions
and each plotted post-eruption JJAS season, Fig. 10) around latitudes
10–15
Importantly, the regions of the most rapid onset, greatest persistence, and
intensity of response included Lake Tana (12
What is certain is that the scale and persistence of the hydroclimatic impacts implied by our modeling for the 168–158 BCE eruption quartet supports, to begin, inferences of poor Nile flooding in 166 and 161 BCE from scattered references in surviving written sources (Bonneau, 1971). These also identify 169 BCE as potentially experiencing poor flooding. This suggests (assuming sufficiently accurate ice-core dating; Sigl et al., 2015) that the eruption quartet (the impacts of which are now better supported and characterized by our modeling) may have compounded the stresses arising from this initial summer of poor flooding, contributing to what is long recognized as a tumultuous decade in Egyptian history. During the 6th Syrian War, Antiochus IV and his Seleukid army invaded Egypt twice. The first invasion occurred in 170 BCE and the second, more serious occupation, in 168 BCE. This takeover might indeed have re-shaped Mediterranean history had it not been averted by self-interested Roman diplomatic intervention (Hölbl, 2001). Internal turmoil continued in Egypt in the 160s and 150s BCE, affecting both the capital (Alexandria) and the countryside. Surviving sources refer, for example, to the experience of “bad times and [people having] been driven to every extremity owing to the price of wheat” in 168 BCE (Bagnall and Derow, 2004, 281–282), and it is known that by the middle of the decade an Egypt-wide agricultural crisis was underway, bringing Ptolemaic officials to near panic (Hölbl, 2001).
Manning et al. (2017) have identified dates of probable revolt onset in Ptolemaic history, with such onset dates identified in 168 and 156 BCE, both coinciding closely with the dates of our eruption quartet. A study of the longevity and geography of these revolts is now of considerable interest. The surviving texts do not tell a complete story, but scattered written references that imply a long persistence of revolt throughout the decade (Veïsse, 2004) are now rendered potentially more explicable given the modeled persistence of reduced temperatures and suppressed Nile summer flooding for more than a decade following the 168 BCE tropical eruption and the three following extratropical eruptions. More precisely, delineating the political, military, economic, and cultural pathways through which any volcanically induced hydroclimatic shock will have propagated is the subject of ongoing efforts to achieve a fuller understanding of the human–environmental entanglements of the 160s BCE. Relatedly, open questions remain as to how directly or indirectly (as per Gao et al., 2021) hydroclimatic shocks may have contributed to the revolts and other societal stresses that feature so prominently in Ptolemaic history, or alternatively (as per White and Pei, 2020), how such shocks may be characterized within a causal schema of necessary and sufficient conditions that might have given rise to revolt and other stresses.
Details to support the results in the paper are available, as the Supplement is provided with the paper. Raw data and codes are available on request from the corresponding author.
The supplement related to this article is available online at:
FL and JGM identified the study period in consultation with the other authors and broader CNH-L-1824770 project team. RS, KT, and ANL designed the model simulations. RS performed the simulations and created the figures in close collaboration with KT, ANL, FL, and JGM. RS wrote the first draft of the paper, and RS and FL led the writing of subsequent drafts. All authors contributed to the interpretation of results and the drafting of the text.
The contact author has declared that none of the authors has any competing interests.
Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
This article is part of the special issue “Interdisciplinary studies of volcanic impacts on climate and society”. It is not associated with a conference.
Ram Singh, Kostas Tsigaridis, Francis Ludlow, and Joseph G. Manning acknowledge support by the National Science Foundation under grant no. CNH-L-1824770. Allegra N. LeGrande acknowledges institutional support from NASA GISS. Resources supporting this work were provided by the NASA High-End Computing (HEC) Program through the NASA Center for Climate Simulation (NCCS) at Goddard Space Flight Center. The authors thank for their input through multiple discussions the project members and collaborators of the CNH-L-1824770 project, “Volcanism, Hydrology and Social Conflict: Lessons from Hellenistic and Roman-Era Egypt and Mesopotamia”. Francis Ludlow acknowledges support from the Trinity Center for Environmental Humanities. This paper benefited from discussion facilitated by the “Volcanic Impacts on Climate and Society” (VICS) Working Group of PAGES.
This research has been supported by the Integrative and Collaborative Education and Research (grant no. CNH-L-1824770).
This paper was edited by Matthew Toohey and reviewed by three anonymous referees.